insurance company benchmark data set gives information on customers. Specifically, it contains 86 variables on product-usage data and socio-demographic data derived from zip area codes. There are 5,822 customers in the training set and another 4,000 in the test set. The data were collected to answer the following questions: Can you predict who will be interested in buying a caravan insurance policy and give an explanation why? Compute the OLS estimates and compare them with those obtained from the following variable-selection algorithms: Forwards Selection, Backwards Selection, Lasso regression, and Ridge regression. What methods performed the best in the prediction?

(The data can be downloaded from https://kdd.ics.uci.edu/databases/tic/tic.html. )